MRI-based Brain Connectivity Methodologies And Clinical Applications

Joint together 100 billion neurons—with 100 trillion connections—the human brain is known as one of the complex system. Nevertheless, functional segregation and functional integration are related to the dialectic between localizationism and connectionism, which dominates ideas about brain’s functional architecture and operational principles. The functional segregation refers that a cortical area is specialized for some aspects of perceptual or motor processing, and this specialization is anatomically segregated within the cortex. However, the functional integration suggests that many neuronal or distributed brain areas collaborate with each other to complete specific behavioral function, which underlies the brain connectivity. These two distinct principles supplement each other. Various approaches to characterizing the functional integration are in terms of functional connectivity, effective connectivity and structural connectivity.Following methodological development, the current work investigated the human brain network on functional magnetic resonance imaging (fMRI) in detail. We aim to develop these methods and translate then into clinical applications for comprehensively and exactly understanding the human brain connectivity network. Three aspects of this dissertation have been put forward:The first part is investigation of effective connectivity network, which is an also methodological development. In this part, we addressed a kernel Granger causality analysis (GCA) to describe nonlinear effective connectivity of the human brain, for availably describing brain dynamics.First, although it is accepted that linear GCA can reveal effective connectivity, the issue of detecting nonlinear connectivity has hitherto not been considered. In the present work, we addressed kernel GCA to describe effective connectivity in real fMRI data of a motor imagery task. Kernel GCA performs linear Granger causality in the feature space of suitable kernel functions, assuming an arbitrary degree of nonlinearity. Our aim is to demonstrate that kernel GCA captures effective couplings not revealed by the linear case.Additionally, for the first time, we combined ICA, a data-driven approach, to characterize resting state networks (RSNs), and conditional GCA to gain information about the causal influences among these RSNs. We focused on evaluating and understanding the possible effective connectivity within RSNs at meso-scale.Thirdly, we aimed to reveal network architectures of effective connectivity brain network at maro-scale. We proposed a multivariate GCA and graph theoretical analysis on large-scale resting-state fMRI recordings. We aimed to found that some brain regions acted as pivotal hubs, either being influenced by or influencing other regions, and thus could be considered as information convergence regions. Furthermore, we also aim to examine that this effective connectivity network has a modular structure and a prominent small-world topological property.In the second part, we devolved a combination of functional and structural connectivity networks to investigate the patients with epilepsy, which is an also neurological disease application. We found that the functional and structural brain network of epilepsy showed aberrant topological attributes, and the functional-structural coupling negatively correlated with duration may associated with the progress of epilepsy.First, little is known about the changes in functional connectivity and in topological properties of functional networks, associated with patients with mesial temporal lobe epilepsy (mTLE). To this end, we constructed large-scale undirected brain networks derived from functional connectivity among brain regions that was measured by temporal correlation. We suggest that the mTLE alterations observed in functional connectivity and topological properties may be used to define tentative disease markers.Secondly, according to disrupted functional connectivity in the default mode network (DMN) in mTLE patients as previously suggested, we combined functional and structural connectivity to study the DMN of those patients. We found that functional and structural connectivity between posterior cingulate cortex and bilateral hippocampus were both decreased. Moreover, combining these two modalities is effective and reliable in MRI data processing, and thus provides us a new way in brain research.Finally, as a global feature of the pathophysiology that widespread brain regions and extensive networks are involved in idiopathic generalized epilepsy (IGE), complex brain network investigation derived from functional and structural connectivity networks based on graph theoretical analysis might be more valuable than local connectional investigations to understand the mechanism of IGE. Importantly, we found deceased functional–structural connectivity network coupling in IGE, and this decoupling was related to duration of the disorder, suggesting that the functional–structural connectivity network coupling may reflect the progress of IGE. Overall, the present study demonstrates for the first time that the IGE is associated with a disrupted topological organization in large-scale brain structural and functional network, opening up new avenues to a better understanding of this disorder.In the third part of this dissertation, we devolved a combination of functional and structural connectivity to investigate alterations on the patients with social anxiety disorder (SAD), which is an also psychiatric disorder application. We found that the SAD showed an aberrant functional and structural brain network, and abnormal effective connectivity network associated with the amygdale.First of all, comparing to patients with SAD and healthy controls, combining spatial independent component analysis (ICA) and brain network analysis, we identified and investigated statistical differences on eight RSNs, such as auditory, visual, somato-motor, dorsal attention, central executive, self-referential, core and default mode networks. Our findings might supply a novel way to look into neuro-pathophysiological mechanisms in SAD patients.In addition, the amygdala is often found to be abnormally recruited in SAD patients. To address this issue, we investigated a network of effective connectivity associated with the amygdala using GCA on resting-state fMRI data at mrico-scale. Our results lend neurobiological support towards SAD.Finally, we integrated voxel-based morphometry, resting-state functional connectivity analysis, and diffusion tensor imaging tractography to investigate brain morphometric, functional, and structural architecture of SAD. For the first time, this work and findings may provide a valuable basis for future studies combining morphometric, functional and anatomical data in the search for a comprehensive understanding of the neural circuitry underlying SAD.